DSP FPGA mixes with the design
In order to raise the algorithm efficiency, the real-time processing pictorial information, this processing system is mixes based on DSP FPGA with the structural design. The service board take FPGA as the processing core, realized the digital video signal real-time imagery processing, DSP has realized the part imagery processing algorithm and the FPGA control logic, and responded the interrupt, realized the data communication and the memory real-time signal.
First, this system requests DSP to be possible to satisfy the algorithm control structure to be complex, the operating speed is high, the addressing is flexible, traffic capacity formidable request. Therefore, our select order cycle is short, the data turnover rate high, traffic capacity is strong, set of instructions function complete DSP. Simultaneously has also considered essential factors and so on DSP power loss and development support environment.
Because transmits from the sounding set the underlying bed A/D signal, its differential value pretreatment algorithm’s data quantity is big, is high to processing speed’s request, but the operation structure is simple, selects 1,000,000 level FPGA to carry on the hardware to realize. Uses DSP FPGA mixes with the hardware system both’s merit to unify together, namely has given dual attention to the speed and the flexibility, also has satisfied the first floor signal processing and the high-level signal processing request. Therefore, very suitable real-time signal processing system.
System construction design
CompactPCI as the PCI main line’s electricity, the software and the industry assembly standard, is now one of newest computer standards. The CompactPCI main line high speed, firm, is reliable, is stable, with the PcI software’s good compatibility, enables it to become the labor to control the domain to be most popular and the general computer interface bus.
CPCI highest transmission speed 528MB/s, the available PCI-X highest transmission speed may reach 1066MB/s at present.
In high speed firm, in reliable stable technology base, this system design has been possible to utilize the customer innate agreement the CPCI back board and the connection series, highly the modular CPCI service board.
Comprehensive handling of traffic platform
The comprehensive handling of traffic platform is refers to the sole platform realizes multi-channel duties and so on signal pretreatments, complex image algorithm, image display, data storage, systems control. Not only this requests the hardware to have the high performance, may carry on the real-time processing, simultaneously, the embedded application environment also requests the volume to be small, the weight is light, the function is strong, the reliability is high.
The complete system constitutes by the following several modules: Box body, power source, back board, A/D pretreatment board and signal processing board and so on.
Box body, power source and back board
The box body uses the standard 19 inch top carriage’s outlook sizes. Inner space: Supports 2U 4 CPCI back boards; Supports 2 3UCPCI power sources. The box body back double mains input connection, the make-and-break type switch (support often opens). After order to respond the power failure, the system restarts request.
3U the CPCI power source supports inserts hotly pulls out; Uses with the department entity intelligent management power source back board; Supports the AC input.
On the back board has 4 6U slots, each slot has 5 plugs: P1, P2, P3, P4, P5. P1, P2 is standard PCI, provides the 5V/3.3V signal environment.
System trough P3, P3, P5 definition according to system board MIC-3369 definition standard.
Expansion slot: P1, P2, P3, P4, P5 uses the penetration long needle, around penetrates, matches the protective covering. P1, P2 designs like this, the first flapper and the back plate may act according to the physical demand to obtain the power supply from the back board. P3, P4, P5 provides the consummation around the signal the route.
In addition, between between 3 expansion slot’s P3, P4, P5 the design is PIN TO the PIN connection.
Such design, constructs for the service board between stores the linear expansion, on the first-level processing module has established physics communication port with under first-level module correspondence.
A/D pretreatment board
Considered the system each service board must process the multi-channel inputs, moreover the project the erection unit request convenience, we design a standard processing module specially for analog input’s signal recuperation: The size is 233.35mm×80mm×1 the trough space. Such holding wire turns on completely behind the box body. Each service board 11 correspondence pretreatment’s digital signal transmits high speed according to the predetermined way through P5 to corresponds on the service board FIFO. The FIFO controller effective transforms the working according to triggering.
Considered the system expansion and trouble shooting’s convenience, the handling of traffic board designs the unification the construction. Thus, user in view of different processing service, so long as changes designs the good software essence, the hardware interface procedure and the user interface does not use the change. Similarly, when trouble clearing, so long as replaces asked that the type the service board then completes.
Handling of traffic board size for 233.35mm×160mm×1 trough space; Supports PICMG 2.1 to insert hotly pulls out the standard.
On the board DSP and FPGA have RAM respectively, uses in data which deposits the handling of traffic process to need.
PMC I/O expands the exhibition board
When practical application in project, the modular system unit usually needs to accept the exterior instruction or through the specific I/O connection output data. We used the PMC card to solve. PMC (PCIMezzanine Card) standard IEEE 1386 have given the mezza-nine module standard. It provided one kind in view to carry the board specification high performance-to-price ratio to realize the I/O function way differently.
The PMC standard is the PCI main line signal reflection to the PJ386 board card. Before the single module size list ((74mm ×149mm) on, breaks out the part to use for to put through I/O, through the sign is P1, P2, P3, the P4 four coupling heads with carries on the board the PCI interconnection.
The user may accept according to the scene request choice standard PMC network interface card with the transmission data. May also defer to needs to have custom-made each kind of serial port I/O connection card, completes the common time-base signal dispatch.
System processing board
MIC-3369 uses the low power loss Pentium-M processor and the optimized Intel*E7501 ICH4R chip set, has supports 64BIT/66MHz the system bus, provided the 3.2GB/s band width, the performance had the extremely good competitive power. MIC-3369 supports the PICMG2.16 standard in the design, the compatible PICMG 2.9 standards, can with the long-distance management system management system platform joint operation.
System application
(far) infrared telemetry system
Develops the infrared digital video signal image formation in (far) the infrared telemetry system the real-time processing system.
When carries on the remote sensing observation and sighting to the great criterion space or the complex terrain, has used many infrared frequency band detectors. Carries on processing according to the different accuracy data, draws up the picture output alphanumeric display terminal, and supplies the mass data high-speed storage to the local hard disk the following use.
System’s principle of work frame graphical representation in Figure 4.
Real-time acquisition system
Have used the industry video processing system in the IC production or the precision component abrasive machining. In such system, the parallel input output’s signaling frequency is not generally high, but the timely request is very strict to the signal processing precision.
We applied the above construction to realize a real-time object detection system, this system’s duty main foot received many location uploads the position quantity sensor, the gradation image which the displacement quantity sensor and the camera output, after pretreatment, code, algorithm processing and target identification, the output result to the alphanumeric display terminal, simultaneously issued the instruction to the control circuit. And, underlying bed processing, its operation data quantity is big, but the operation structure is regular, is suitable for to carry on the pure hardware with FPGA to realize; But algorithm processing and the target identification and so on high-level imagery processing, needs to use many kinds of agreement structures, realizes with the DSP programming. System’s schematic drawing as shown in Figure 5.
Conclusion
Uses architecture which this article said that can realize the high density redundant reliable system rapidly in an open modular platform. It has the following characteristic:
1. low MTTR, high usability;
2. the system disposition is flexible and the promotion, the maintenance are convenient;
3. the convenience implementation system has custom-made, reduces the research and development cost;
4. realizes the exclusive application rapidly, enhances the competitive advantage.
